Skip to content

Search the site

    Methods · 12 October 2026 · 5 min read

    Same handle, different person: the trouble with username matches

    Username search tools tell you a page exists, not who owns it. Here is how to weigh handle matches, spot false positives and false negatives, and avoid naming the wrong person.

    A username is a name tag that anyone can wear. Two people can pick the same one on different sites, and one person can wear many. That simple fact is the root of a common investigative mistake: treating a matching handle as a matching human.

    Imagine a community group that finds a critical post from an unfamiliar account. Someone types the account’s handle into a username search tool, six sites light up, and a theory is born. (This is an illustration, not a documented incident.) The theory might be right. The tool has not shown that it is.

    What a username search actually checks

    Username search tools, sometimes called name checkers, work by trying a handle against a long list of sites and noting whether a profile page appears at the expected address. A result of “taken” means a page exists for that string on that site. That is all it means.

    It does not say the account is active, that it belongs to the person you are curious about, or that two “taken” results belong to the same owner.

    The tools also get things wrong in both directions, and their own communities say so. The creator of the WhatsMyName project has said the project began because existing checkers produced too many false positives. Sherlock’s public issue tracker shows maintainers removing a site over false positives, and users reporting results that did not match what they expected. Sites change how they respond, and a detection rule that worked last year can start flagging every handle.

    Two kinds of error follow:

    • False positive: the tool says “taken” for a page that is empty, unrelated or belongs to a stranger.
    • False negative: the person has an account, but under a slightly different handle, so the tool never finds it. A hyphen or an underscore is enough.

    Not every handle carries the same weight

    Researchers at INRIA studied how well usernames alone can link profiles across services. Their central idea, in a 2011 paper, is that the chance two matching usernames belong to the same person depends on how unusual the string is. A short, common handle is weak evidence. A rare string is much stronger. They also found that people tend to settle on a small set of related usernames and reuse them widely, which is why variations such as a hyphenated copy of a handle often belong to the same person.

    Both halves matter. A match on a common handle proves little, and a near match on a rare handle is worth following up.

    Three kinds of evidence

    When you compare accounts, sort what you find into three groups.

    1. The handle itself. Weak to moderate, depending on how unusual it is.

    2. Soft matches. Same city, similar interests, posting at similar times of day, a similar writing style. Each of these is shared by thousands of people, so each is weak on its own. They add up only when several agree, and only if they are independent of each other.

    3. Hard links. An account that points to another, ideally in both directions. A shared piece of content that appears first on one account. A contact address on a domain the other account’s owner also controls. Hard links are your strongest evidence, but check their direction. A blogger can link to any account they like, so a one-way link proves less than a two-way link. Also remember that details such as an email address attached to a code commit are self-declared and can be set to anything.

    Public details can be copied too. A distinctive sign-off phrase is useful, but anyone who has read the original can reuse it.

    Keep a “different person” column

    Most people build a case by collecting reasons an account matches. Build a second list as well: reasons it might belong to someone else. A different time zone, a different writing style, an account older than the thing you are investigating, or evidence that the handle was already in use by an unrelated person all belong there. The strength of a conclusion shows in how well it survives that second list.

    What it costs to get it wrong

    Online sleuthing has a record of naming the wrong person. After the 2013 Boston Marathon bombing, amateur investigators on social media wrongly named a missing student as a suspect. His family, who had been searching for him, ended up dealing with the fallout, and Reddit’s general manager later apologised publicly. That case turned on faces and rumours, not usernames, but the lesson carries over. Confident, shared, wrong conclusions travel faster than corrections.

    Weigh your confidence against the cost of being wrong, and write down how sure you are.

    A short method

    1. Write the claim as a question: is account A the same person as account B?
    2. List every candidate account with its handle and the reason it is on your list.
    3. For each, record evidence for and against, and label each item as handle, soft match or hard link.
    4. Note which pieces of evidence could have been copied or faked.
    5. Give each account one verdict: same person, probably, cannot determine, or probably different.
    6. Choose a proportionate action. Often the safest step is not to accuse anyone, but to ask the person directly.

    A defensive note

    The same research explains why reusing one handle everywhere makes you easy to trace. If you want to keep parts of your online life separate, use different handles for them, and run your own handles through a name checker to see what a stranger would see.

    Try the exercise

    Exercise 002, Same handle, different person gives you a fictional community forum, a name-check result and seven profile cards. Decide which accounts belong to the same person and how sure you are. Everything in it is invented, so nobody can be harmed by your guesses.

    Sources

    Related posts

    Back to all posts